Delayed Diagnosis of Rare Diseases Takes Staggering Economic Toll

The economic impact of a delayed diagnosis of a rare disease can result in up to $517,000 in avoidable costs per patient, according to a recent analysis from the Everylife Foundation for Rare Diseases and the Lewin Group. On average, the report found, it takes more than six years and nearly 17 doctor visits, hospitalizations, and other health-related trips, to receive a rare disease diagnosis after symptoms begin. Shortening the diagnostic odyssey saves money for individuals, caregivers, and the healthcare system, while improving health outcomes. We spoke to Annie Kennedy, chief of policy, advocacy, and patient engagement for the Everylife Foundation, about the diagnostic odyssey, the economic toll it takes, and what can be done to shorten the time to a diagnosis.

Annie Kennedy: Thanks so much for having me. I’m delighted to be here.

Daniel Levine: We’re going to talk about rare disease diagnosis, the high cost of delayed diagnosis, and the EveryLife Foundation’s recent study in conjunction with the Lewin Group about this. Let’s start with the headline. What did the study find?

Annie Kennedy: So, thanks so much. As you recall, and I’m sure many listening will recall that we, in 2022, published our National Economic Burden of Rare Disease Study where we found that in 2019 the cost of living with a rare disease was over, or close to, a trillion dollars. And what was important about that study was that we unveiled what many of us in the rare disease community know—that the overall large proportion of those costs are the costs being shouldered directly by families and society. So, of those costs, about 60 percent of those were those indirect or non-medical costs, those costs that are coming out of pocket for families. But in that same study, we found that the diagnostic odyssey was about six years and included 17 medical visits, so we wanted to go below the surface and learn more about that diagnostic odyssey. And in this study we found that in the diseases that we explored, and we’ll talk more about how we got to that, the avoidable costs ranged from $86,000 to $517,000.

Daniel Levine: In terms of the total cost, how much of that was medical costs versus indirect cost?

Annie Kennedy: So, the lion’s share of that, the large proportion of that, are direct medical costs. We did a cost estimate around productivity losses, but that’s a super conservative estimate and that’s probably less than 5 percent of those costs. So, the large percentage of these costs that we’re talking about are those direct medical costs that are a part of the diagnostic odyssey that is completely avoidable in the diseases that we’re talking about here today.

Daniel Levine: Well, maybe we can take a step back. How do you go about calculating those costs? What data sources did you use and what was included in that calculation?

Annie Kennedy: Sure. We used the Medicaid dataset, we used the Medicare dataset, and we used a very large commercial dataset through Optum Health UnitedHealthcare. So, we were looking at more than 30 million lives—so a very large representative portion of the community. And as you recall in the first study, we were really trying to look at the broad estimate of rare diseases. So, in the first study we looked at 379 rare diseases. For this study, we really needed to do a deeper dive into specific rare diseases. So, for this study we were looking at seven specific rare diseases. And to look at these diseases, we were really looking at the ICD code or the International Classification of Diseases, the identifier within those data sets for each of those diseases.

Daniel Levine: My sense is that the numbers here are actually conservative. How limited were you by the data you were able to document?

Annie Kennedy: Doing this work is really complicated. In rare disease, one of the limitations is that many rare diseases don’t have ICD codes at all. So that’s one of the limiting factors in doing any health economics work overall. We selected diseases that we knew had ICD codes, but even within the diseases–so the seven diseases we looked at, I should probably say—we chose seven diseases, some of which were pediatric onset, some of which are adult onset because of course we care about the diagnostic odyssey for all those. In the rare disease community, we chose some diseases that have newborn screenings, some that do not, and we wanted to make sure all of them had an ICD code—that we would be able to look for this data within direct-cost data sets. Within these data sets we needed that code to do so. But within those diseases, some of these diseases had not had an ICD code for very long, so that means that we may not have data for the entire time period that we were looking. The other thing is that, for example, one of the diseases that we looked at was Pompe disease, and there’s one ICD code, even though there are different forms of Pompe, there are different subtypes. And so that means that you have to apply different algorithms to try and tease out the infantile onset Pompe from late onset Pompe. That means that you are impacting the strength of the data. Another disease that we included in a study was Duchenne muscular dystrophy, and again, there’s one ICD code for both Duchenne muscular dystrophy and Becker muscular dystrophy because those are two disorders that are on the same spectrum, but share one ICD code. So, we know that when we’re looking at this data, we have a conservative estimate of the findings because we are probably mixing different phenotypes, different subtypes of these disorders when we look at this data.

Daniel Levine: The study used real world data to look at healthcare utilization relative to the timing of diagnosis across seven diseases as you’ve mentioned. What were the sources of the data you used?

Annie Kennedy: Yeah, so again, we used those three main commercial data sets. The other thing, though, that we did is we were able to leverage the data from the survey that our patient community powered in 2020 when we first conducted our early work. So you’ll recall that our community helped develop a survey that we then disseminated through our patient community partners in the summer of 2020. And that survey served as the basis for that indirect and non-medical data that we used in the 2022 publication. We were able to then create additional data for this study to do that productivity loss analysis that we talked about. So, it’s really important to us that when we talk about the diagnostic odyssey, we also think about what’s happening when we’re having this long six to, for some diseases we found much longer than 6-year odyssey of medical visits, emergency room visits, hospitalizations, ICU stays. We all recognize that that means that there are caregivers involved, people missing work, individuals not being able to go to their jobs while they’re staying home with other family members who are other children who are at home. And we needed to create a calculation that was accounting for that time. So, we created an algorithm that correlated for every inpatient day, a day missed work, for every outpatient or emergency room visit, a half a day missed work, and for every physician office visit, a quarter of a day missed work. One of the things—to your earlier question about this being very conservative—we know that that is a very conservative calculation when you talk about rare disease. We know that for many in the rare disease community going to see your inpatient specialist, your specialist is not one day of missed work. Many people have to travel a great distance to see their physician, but we had to anchor this algorithm, this methodology, to something published. And there hasn’t been a lot of published data in general, and there certainly hasn’t been anything published around rare disease. So we had to start somewhere. So we anchored it to something that had been established in premature infants and then created this methodology for rare diseases. But this will skew conservative because we’re starting someplace.

Daniel Levine: One of the reasons I think the numbers are probably conservative is because in your selection of diseases, one of the criteria was the disease had to have an ICD code, which implies it’s relatively diagnosable compared to rare diseases broadly, at least in my mind. I am curious, what other criteria did you use in selecting the diseases, and what were the seven diseases in the study?

Annie Kennedy: Yeah, that’s a great question. So I’ll answer that and then I want to make another comment about your observation, which I think is spot on. So as I said, this was an important start for us. We, on purpose, wanted to, if whenever we had to skew one direction, we wanted to skew conservative because we as a community are using this evidence for policy work. And so it’s important that it be bulletproof and defensible. We wanted to make sure that we were getting data about the diagnostic odyssey for both our pediatric community and our adult community. We do a lot of work certainly in the newborn screening space, but as we all know, a large percentage of rare diseases have onsets much later than the pediatric time period. So that was really important to us. We also wanted to make sure that we were looking at the impact of newborn screening on the economics of rare disease. One of the things that is a really high bar for getting a condition added to the recommended uniform screening panel, or the federal panel, is how much that screen costs. So it was important to us to include a few conditions that had already been added to the federal panel so that we could look to see what the avoidable cost spend was, what were the cost savings, and even for those conditions that are on the federal panel, those conditions are not necessarily all being screened for in all 50 states. So could we look at having some data that would help embolden the advocacy efforts of our communities that are working to have conditions added in all 50 states? So, for that reason, of the seven conditions, we have five pediatric [and] two adult conditions, and of the five pediatric, three are conditions that are already on the RUSP. So, our five pediatric conditions were ALD, adrenal leukodystrophy, Pompe disease, and SCID, and those were the three conditions that are already on the RUSP. And then the two pediatric onset conditions that are not yet on the RUSP were fragile X and Duchenne muscular dystrophy. And then we also had, for the more traditionally adult onset conditions, Wilson disease and generalized myasthenia gravis. So those were the seven conditions that we looked at. The other thing I just wanted to comment on from your earlier comment about why you think this may be skewing conservative, just another comment about that is in addition to these conditions, [which] all had I C D codes, when we talk about the diagnostic odyssey, it almost presumes that when we look back, we know when the diagnostic odyssey starts. And I think for many of us, the start date isn’t typically that first medical appointment or intervention or the time point the dry blood spot was taken. When you have a diagnostic odyssey, oftentimes it’s in retrospect, you look back and think, “oh, these other things were connected also to that diagnosis.” And so there could be so many other medical appointments or conditions that ended up being related to that final diagnosis that might never get counted as a part of those costs because we don’t necessarily connect them until later. So that’s another reason why when we do this work, it’s not necessarily clean work because when we do this patient journey mapping, we learn later about other events and symptoms that may have been ultimately related to the underlying cause.

Daniel Levine: You spoke to this a little, but one of the things that may be confusing is that you’ve got three conditions here that are actually part of the universal newborn screening. Why is there a diagnostic odyssey if these are tested at birth? Is that because there’s only a handful of states that are actually incorporating that?

Annie Kennedy: Yeah, so not all states. So once a condition is added to the federal panel, that is a recommendation. It’s called the Recommended Uniform Screening Panel (RUSP). And then there is still a state by state process by which conditions are then considered for implementation in the states. And that process can take sometimes up to a decade for implementation from state to state. And so it’s really critically important that we have economic evidence for communities as they think about how to shorten that timeline and advocate for conditions to be added to the states. That’s oftentimes an economic consideration for states when they think about which conditions to add and how to make that investment for adding a condition. So it was really important for us to have this data so that we could show that, well, what public health labs are considering is the cost of adding a public health screen and screening a whole population, which could be a matter of dollars. Whether we like it or not, we’re already spending the money because we’re spending the money on the diagnostic odyssey and we’re spending the money to the tune of hundreds of thousands of dollars on people’s avoidable diagnostic odyssey, which yields irreversible disease progression. And what we’re causing is for people to miss oftentimes that optimal treatment window.

Daniel Levine: Let me ask you a more basic question. For listeners who may not live in this world, what makes diagnosing someone with a rare disease so difficult to begin with?

Annie Kennedy: I mean, it shouldn’t be. I’m just going to say that from the outset, we have the technology to do better. We have the technology that this shouldn’t be so complicated and so difficult, but one of the things that happens, I think we all know that the zebra is our mascot, and one of the reasons is because providers and physicians are really taught that rare diseases are rare and that when you think of symptoms related to rare diseases, you shouldn’t actually consider a rare disease. You should think of more obvious things. And so physicians are actually taught when you hear hooves, think of horses, not zebras, don’t think about something rare. And so it does make it really hard. Another thing is that oftentimes people have seizures, and so what you think about more are other types of developmental disabilities, but you don’t think about once you’re diagnosed with the seizure disorder doing the genetic screen to see what could be the underlying cause of that disorder. So again, if we were to implement tools through genetics and through the technologies that we have available, we would find that underlying disorder much more quickly. It shouldn’t be so hard. And then going back to the ICD code issue you raised before, an ICD code is an identifier, it’s like a hashtag. It helps you see issues in the medical system. It connects data, it connects your medical records. In the rare disease community, we have an estimated about 10,000 rare diseases. We only have about a little over 500 ICD codes. There are ICD codes for getting stitches. There are ICD codes for high blood pressure. You have an ICD code for pregnancy–which means if I go to the ER and I get my finger sutured up, there’s an ICD code for that intervention. And so, then when I go to my primary care doctor and she looks at my records, she knows that I got my finger sutured a couple of weeks ago, That follows me. But if you have a condition that doesn’t have an ICD code, you’re not being tracked in the medical system. And so if you’re going to multiple providers to seek your care, they can’t put the pieces of the puzzle together really quickly. They can’t see the types of care that you’re getting and they can’t connect it to the underlying cause very quickly. And so it makes it harder to get to that diagnosis faster because we’re not tracking all of those symptoms in one place.

Daniel Levine: We often hear numbers about the length of the diagnostic odyssey. One of the problems with these numbers is that they’re often referred to as if they’re an absolute. In reality, these are usually based on small studies and dependent on the diseases included in that study. What do we know about the length of the diagnostic odyssey and how this can vary not only by disease but by geography?

Annie Kennedy: Yeah, so it definitely varies by disease. We also know that if you’re far from medical specialty centers, it can take a lot longer. We know that a lot of times when you’re in the center of the country, you may not have as much access to specialty care. And we know that depending on who your health insurance is, there may be barriers to getting out-of-state care. So, in that first survey that we conducted, one of the questions we asked was not even just how long was the diagnostic odyssey, but how many medical interventions and providers were included, and then further asked of that, what was the average number of out-of-state trips were included? And we did some calculations in that first study and in this study around the mean number of out-of-state trips for both pediatrics and adults. And while that data is incredibly compelling and important for us to see, it also supposes that you have health insurance that allows you to cross state lines in rare disease. We have a highly Medicaid-eligible population, which is terrific because Medicaid provides a lot of really important coverage and reimbursement. But one of the limitations of Medicaid is that oftentimes you cannot get reimbursed across state lines. So, if your health insurance comes through Medicaid, but the specialist you need to see to get to a diagnosis is outside of your state, that will serve as a barrier to getting a diagnosis. That becomes a huge problem. And that will definitely prolong your diagnostic odyssey if you can’t get to the healthcare specialist that you need.

Daniel Levine: One of the interesting things your study did was to take a look at the number of healthcare events before diagnosis by the length of the diagnostic odyssey, and there’s a compelling chart that breaks this down within the study. What did you find in terms of variance of time of diagnosis and the cost implications of that?

Annie Kennedy: Yeah, so we did a couple things. We also sliced and diced it by the age of the individual who was being diagnosed. But to your question, around the number of healthcare events by time to diagnosis, we looked at whether or not, if you were diagnosed, the length of your diagnosis was less than two years, the number of out-of-state trips that you would have to take in children, and then sort of going above five years. And we saw everything ranging from if your diagnostic odyssey was less than two years, you had an average about a one and a half to two out-of-state trips versus if your diagnostic odyssey went more than five years, you were going out of state more than five times to get to a diagnosis in peds. In adults, for under two years, it was very similar. It was about one and a half trips out of state, whereas the diagnostic odyssey in adults often went more than 12 years. It took adults much longer to get a diagnosis, and that often required closer to four out-of-state trips. So we really see that as that diagnostic odyssey gets much, much longer, it requires a lot more out-of-state travel and it significantly exacerbates the costs to get a diagnosis, not just because of the productivity losses, but then the spend. So we were able to really correlate that to the direct costs, whether we were looking at inpatient visits, ER admissions, ICU stays, the outpatient visits, we were able to really break down the data of where that money was being spent. Again, indirect costs, and it’s really important to highlight that these are avoidable costs when we look at the difference between a timely diagnosis and a delayed diagnosis.

Daniel Levine: Newborn screening in the United States seems not only fragmented but broken. What’s the state of newborn screening in the United States and how well served are we by it as it stands today?

Annie Kennedy: Well, I mean I think it’s important to first say that newborn screening really is a critical foundation for the U.S. and it is really thought to be one of our most successful public health programs. That said, it was not built to withstand the innovation that we have today. So when I think this system was stood up, I don’t know that anyone conceived of the number of therapeutic opportunities that we would have today or the pace of innovation that we have today. So we have this extraordinary bottleneck right now. We have so many more therapeutic and intervention opportunities than we’re able to identify patients for. And so, I think the other thing that’s really important is that the criteria that’s used right now by the advisory committee that makes decisions about what conditions get screened for, at least in our view at the EveryLife Foundation, is probably not the criteria that we necessarily need to be using. And there’s a lot of gray area around that criteria right now. We talk about intervention very broadly and very traditionally. Interventions have often been thought about as whether or not there’s an FDA approved treatment that would warrant screening. But as we all know, life-changing, life altering interventions often do not come in the form of FDA approved treatments. They come in other forms. And for communities with rare diseases, often knowledge itself is power, and the ability to be connected to communities, and be connected to appropriate care, and just be off that diagnostic odyssey sooner is reason itself to be offered an opportunity to be diagnosed at birth. And so we really think that we need to be innovating and modernizing this newborn screening system and ensuring that the therapies and interventions that are available are being offered at the earliest moment possible to infants and children, and that we have an updated modernized newborn screening system that is really set up for the families and babies who are being born today.

Daniel Levine: I think the point you just made is so important because the bioethical argument against testing and revealing the presence of a gene for disease for patients where there’s no treatment goes against testing, but it’s so contrary to the experience of people in the rare disease community, what they’re able to do once they have a diagnosis. Do the bioethicists have this wrong?

Annie Kennedy: I think it comes down to [is] it’s oftentimes a personal decision for families. I think we have to really expand how we do consent for families. And when we do consent, I think this is going to take a lot more involvement of the obstetrics community. I think right now that’s a community that is not as involved in newborn screening, and a lot of this needs to be done before somebody’s in a birthing center, in a birthing hospital so that these conversations are happening earlier as well. And I think we also need to think about what types of information and what types of conditions we’re screening for and what that criteria is. I think it’s important to think about the fact that right now we spend a lot of time with the secretary’s advisory committee thinking about conditions that could be screened for that would affect children in that newborn period or that might affect them under the age of two. But really the charter of that committee is it’s the advisory committee of heritable disorders in newborns and children. And so we really need to think about what is that threshold for identification and are we really talking about conditions that can be identified and treated and that have meaningful interventions that would be otherwise devastating or life altering under the age of 18? And that’s already the charter, but it’s not really what we talk about in that committee. So I don’t know that I would say they have it wrong. We’re just not actually having the right conversations, which might be maybe a political way to skirt your answer, Danny, but I do think we are pushing really hard that we need to have very different conversations. And the other issue is we don’t have enough patient community members in the center of these conversations. These are oftentimes largely conversations and decisions that are being had by the bioethicist, and it’s not the bioethicist who should all be making the decisions. We really need the experienced patient community members who’ve had this lived experience who should be at the center of the conversations helping inform these decisions and these policies.

Daniel Levine: There is growing evidence of the power of whole genome sequencing as a newborn screening tool. I think the United States here may be a little behind the United Kingdom, but what do you think it will take to get there?

Annie Kennedy: I mean, absolutely. We absolutely believe that next gen sequencing is a super powerful tool and has a lot of potential, and sequencing, no doubt, has potential to have a really powerful impact. It is still in early days, and there have been some really exciting and important pilots, and those pilots have really shown that it really has the ability to unleash some potential to promote the timely diagnosis associated with newborn screening. I think one of the challenges is right now, in the U.S., our newborn screening system is built to support metabolic screening and not next gen sequencing. So, we really have to think about how we will innovate the existing system to build both, and network to support both types of testing and analysis. And we also have more to understand about how next gen sequencing works in the general population and how we’re going to store data and how we will implement this within our broader existing newborn screening system. So, I think we are really cautiously optimistic and excited about the potential and eager and grateful to be a part of these conversations, but understand that we certainly have more to do. And I will flag that we put out a paper just a couple weeks ago around pioneering modernization in our newborn screening system, and the inclusion of next gen sequencing within our newborn screening system is one of the areas that we have flagged as a priority.

Daniel Levine: The report is The Cost of Delayed Diagnosis in Rare Disease. It can be downloaded for free from the EveryLife Foundation for Rare Diseases [email protected]. Annie Kennedy, chief of policy advocacy and patient engagement for the EveryLife Foundation. Annie, thanks as always.

Annie Kennedy: Thank you so much for having me.

This transcript has been edited for clarity and readability.